OldSnap Account Download

Image to Excel

Paste a screenshot of a table, fix any cell, then download it as .xlsx or .csv.

PNG, JPG or WebP · up to 10 MB

Your image is only used to read the table. It isn’t stored on our server. It goes through OldSnap’s server to DeepSeek’s vision model. More

How to convert an image to Excel

  1. 1

    Add the image

    Drop it on the box at the top, click to pick a file, or press Ctrl V to paste a screenshot. PNG, JPG and WebP up to 10 MB work.

  2. 2

    Check the cells

    The table comes back as a grid. Click any cell to fix it, and add or delete rows if something got split or merged.

  3. 3

    Download or copy

    Save an .xlsx or a .csv, or copy the table and paste it into Excel or Google Sheets. It lands as cells, not one long line.

Windows

Screenshot to Excel on Windows

If you pull tables out of screenshots every week, the OldSnap desktop app is quicker than uploading files here.

Press Ctrl 1 and drag over the table. Press O to read the text. When it lines up like a table, a Copy as table button shows up. Click it and paste into Excel or WPS. Each value lands in its own cell.

The text is read offline by a built-in OCR model, so nothing is uploaded and there’s no daily limit. Capturing, text recognition and Copy as table are free.

Download for Windows What else it does

Windows 10 / 11, 64-bit, no admin rights needed.

Works with

Anything that is a table in a picture. These come up most.

  • Invoices and receipts

    Line items, quantities and totals. Amounts keep their currency signs and separators.

  • Quotes and price lists

    Supplier quotes and rate cards that arrive as images or in a chat.

  • Admin dashboard reports

    Tables from back-office pages that have no export button.

  • PDF screenshots

    When copying out of a PDF scrambles the columns, screenshot the table instead.

  • Chinese, Japanese and Korean text

    Characters come out as written. The .csv is saved with a BOM so Excel opens it without garbled text.

Limitations

It reads the image with an AI model. It gets things wrong sometimes, and these are the usual places.

  • Handwriting is often misread. Printed text works much better.
  • Merged cells may end up in the wrong column or be left empty. If the model notices them, it says so above the grid.
  • Headers with two or more levels get flattened into plain rows. You may have to fix the top rows by hand.
  • Blurry or very low-resolution images. If the text is hard for you to read, expect mistakes.
  • Check important numbers. A misread digit looks just like a right one. Compare totals with the image before you send them anywhere.
  • Several tables in one image may end up in one grid. Crop to one table for a cleaner result.

FAQ

Is it free? Is there a limit?
Yes, it’s free and there’s no sign-up. Each IP address gets 3 conversions a day, and the count resets at midnight UTC. Reads that fail don’t count. On a very busy day the tool can hit its overall daily cap and ask you to come back tomorrow.
Is my image stored?
No. Your image is only used to read the table. It isn’t stored on our server. It goes from your browser to OldSnap’s server, which passes it to DeepSeek’s vision model and sends the cells back to you. The server keeps a usage record with the time, a one-way hash of your IP address, and the token count and cost. Not the image, and not the text. DeepSeek handles the request under its own privacy policy. Details are in the privacy policy.
Which formats and sizes work?
PNG, JPG and WebP, up to 10 MB. Before upload, your browser shrinks the image to 1600 pixels on the longest side, which is enough for normal screenshots. If a very long table comes out wrong, crop it into two parts and convert them one at a time.
How accurate is it, and which languages does it read?
It does best on sharp screenshots of typed text. It reads English, Chinese, Japanese, Korean and other printed scripts, and keeps numbers the way they are written. It can still misread a digit or put a value in the wrong column, so check the numbers that matter before you use them.
How is this different from the OldSnap desktop app?
The desktop app reads text offline with its built-in OCR model and turns text that lines up like a table into cells. Nothing is uploaded and there’s no daily limit, but it runs on Windows only. This page uses an AI vision model instead, so it works on any device, but it needs an upload and you get 3 a day.